基于机器视觉的高速公路车辆行驶速度与区间状态研判方案研究

Research on the Scheme of Expressway Vehicle Tracking and Abnormal State Recognition Based on Machine Vision

摘要: 高速公路具有车辆行驶速度高、车流量大和行驶环境复杂多变等诸多特点，传统的基于车辆车牌或单点摄像头识别的监测手段已难以满足对全路域车辆运行状态进行实时动态掌控的需求，尤其是在匝道出入口、高峰拥堵及夜间等复杂等环境下，存在识别准确率低、运行动态无法连续拼接、异常状态发现滞后等问题。本文依托S6济潍高速K35~K51区间沿线布置的视频监控装置，基于机器视觉及时空轨迹推断技术，构建了高速公路区间全路段车辆跟踪与异常状态识别技术方案，以实现对进入车辆全路段智能感知。基于布置在S6济潍高速K38+000和K38+510两处摄像头所获取的视频图像数据为样本，进行了方案有效性的初步检验，验证了其可行性与有效性。本文所构建的技术方案为高速公路全路段车辆跟踪与异常状态识别提供了技术支撑。

Abstract: Expressway has many characteristics such as high vehicle speeds, large traffic volumes, and complex and changing driving environments. The traditional monitoring methods based on vehicle license plate or single point camera recognition are no longer able to meet the demand for real-time dynamic control of vehicle operating status throughout the entire road area, especially in complex environments such as ramp entrances and exits, peak congestion, and nighttime. There are problems such as low recognition accuracy, inability to continuously connect operating dynamics, and lagging detection of abnormal states. This article relies on the video surveillance devices installed along the K35~K51 section of the S6 Ji-Wei Expressway, and based on machine vision and spatiotemporal trajectory inference technology, constructs a scheme for vehicle tracking and abnormal state recognition throughout the entire section of the expressway, thus to achieve intelligent sensing of vehicles since entering the entire section. Based on the video image data obtained from the cameras deployed at K38+000 and K38+510 on the S6 Ji-Wei Expressway as samples, a preliminary test of the effectiveness of the scheme was conducted to verify the scheme’s feasibility and effectiveness. The technical solution constructed in this article provides technical support for vehicle tracking and abnormal state recognition throughout the entire expressway section.